Mercurial > hg > anteater

changeset 3:ae96e4bc7fb2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
save found species to analysis files
author jdamerow
date Mon, 22 Oct 2012 14:21:14 -0700
parents 1c2b4f5e2c05
children dcc35f89dce3
files analysis/00-18565.xml analysis/01-14522.xml analysis/01-19062.xml analysis/2010-23822.xml src/de/mpiwg/anteater/AnteaterController.java src/de/mpiwg/anteater/species/common/CommonNameFindController.java src/de/mpiwg/anteater/species/common/impl/LinnaeusNameFinder.java src/de/mpiwg/anteater/text/TextInformation.java src/de/mpiwg/anteater/xml/impl/AnalysisXMLManager.java src/de/mpiwg/anteater/xml/impl/templates/analysisFile.xml
diffstat 10 files changed, 165 insertions(+), 7 deletions(-) [+]
line wrap: on
line diff
--- a/analysis/00-18565.xml	Mon Oct 22 13:46:54 2012 -0700
+++ b/analysis/00-18565.xml	Mon Oct 22 14:21:14 2012 -0700
@@ -85,6 +85,8 @@
       </result>
     </supplementary_information>
   </scientificNames>
+  <commonNames>
+  </commonNames>
   <places>
     <summaries>
       <contentlocation xmlns="http://wherein.yahooapis.com/v1/schema" xmlns:yahoo="http://www.yahooapis.com/v1/base.rng" xml:lang="en">
--- a/analysis/01-14522.xml	Mon Oct 22 13:46:54 2012 -0700
+++ b/analysis/01-14522.xml	Mon Oct 22 14:21:14 2012 -0700
@@ -81,6 +81,22 @@
       </result>
     </supplementary_information>
   </scientificNames>
+  <commonNames>
+    <summaries>
+      <linnaeus />
+      <linnaeus />
+    </summaries>
+    <supplementary_information>
+      <linnaeus>
+        <species id="species:ncbi:34886" start="1651" end="1668" text="Steller sea lions" />
+        <species id="species:ncbi:34886" start="1671" end="1689" text="Eumetopias jubatus" />
+        <species id="species:ncbi:34886" start="2561" end="2577" text="Steller Sea Lion" />
+        <species id="species:ncbi:34886" start="7324" end="7341" text="Steller sea lions" />
+        <species id="species:ncbi:34886" start="7344" end="7362" text="Eumetopias jubatus" />
+        <species id="species:ncbi:34886" start="8234" end="8250" text="Steller Sea Lion" />
+      </linnaeus>
+    </supplementary_information>
+  </commonNames>
   <places>
     <summaries>
       <contentlocation xmlns="http://wherein.yahooapis.com/v1/schema" xmlns:yahoo="http://www.yahooapis.com/v1/base.rng" xml:lang="en">
--- a/analysis/01-19062.xml	Mon Oct 22 13:46:54 2012 -0700
+++ b/analysis/01-19062.xml	Mon Oct 22 14:21:14 2012 -0700
@@ -89,6 +89,29 @@
       </result>
     </supplementary_information>
   </scientificNames>
+  <commonNames>
+    <summaries>
+      <linnaeus />
+    </summaries>
+    <supplementary_information>
+      <linnaeus>
+        <species id="species:ncbi:34886" start="755" end="772" text="Steller sea lions" />
+        <species id="species:ncbi:34886" start="868" end="885" text="Steller sea lions" />
+        <species id="species:ncbi:9742" start="2299" end="2315" text="harbor porpoises" />
+        <species id="species:ncbi:9742" start="2318" end="2335" text="Phocoena phocoena" />
+        <species id="species:ncbi:27606" start="2934" end="2961" text="North Atlantic right whales" />
+        <species id="species:ncbi:27606" start="2964" end="2983" text="Eubalaena glacialis" />
+        <species id="species:ncbi:27606" start="3241" end="3268" text="North Atlantic right whales" />
+        <species id="species:ncbi:34886" start="6627" end="6644" text="Steller sea lions" />
+        <species id="species:ncbi:34886" start="6740" end="6757" text="Steller sea lions" />
+        <species id="species:ncbi:9742" start="8171" end="8187" text="harbor porpoises" />
+        <species id="species:ncbi:9742" start="8190" end="8207" text="Phocoena phocoena" />
+        <species id="species:ncbi:27606" start="8806" end="8833" text="North Atlantic right whales" />
+        <species id="species:ncbi:27606" start="8836" end="8855" text="Eubalaena glacialis" />
+        <species id="species:ncbi:27606" start="9113" end="9140" text="North Atlantic right whales" />
+      </linnaeus>
+    </supplementary_information>
+  </commonNames>
   <places>
     <summaries>
       <contentlocation xmlns="http://wherein.yahooapis.com/v1/schema" xmlns:yahoo="http://www.yahooapis.com/v1/base.rng" xml:lang="en">
--- a/analysis/2010-23822.xml	Mon Oct 22 13:46:54 2012 -0700
+++ b/analysis/2010-23822.xml	Mon Oct 22 14:21:14 2012 -0700
@@ -132,6 +132,44 @@
       </result>
     </supplementary_information>
   </scientificNames>
+  <commonNames>
+    <summaries>
+      <linnaeus>
+        <species id="species:ncbi:36500" start="27" end="34" text="S. Fish" />
+        <species id="species:ncbi:36500" start="494" end="501" text="S. Fish" />
+      </linnaeus>
+    </summaries>
+    <supplementary_information>
+      <linnaeus>
+        <species id="species:ncbi:30601" start="3169" end="3190" text="golden-crowned sifaka" />
+        <species id="species:ncbi:30601" start="3193" end="3216" text="Propithecus tattersalli" />
+        <species id="species:ncbi:48399" start="3525" end="3543" text="American crocodile" />
+        <species id="species:ncbi:48399" start="3546" end="3563" text="Crocodylus acutus" />
+        <species id="species:ncbi:184237" start="3571" end="3590" text="Morelet's crocodile" />
+        <species id="species:ncbi:37171" start="3901" end="3909" text="bontebok" />
+        <species id="species:ncbi:36500" start="4280" end="4287" text="S. Fish" />
+        <species id="species:ncbi:29073" start="4445" end="4456" text="polar bears" />
+        <species id="species:ncbi:29073" start="4459" end="4474" text="Ursus maritimus" />
+        <species id="species:ncbi:9708" start="4773" end="4787" text="Pacific walrus" />
+        <species id="species:ncbi:9708" start="4791" end="4818" text="Odobenus rosmarus divergens" />
+        <species id="species:ncbi:9778" start="5132" end="5148" text="Florida manatees" />
+        <species id="species:ncbi:9778" start="5151" end="5169" text="Trichechus manatus" />
+        <species id="species:ncbi:30601" start="8687" end="8708" text="golden-crowned sifaka" />
+        <species id="species:ncbi:30601" start="8711" end="8734" text="Propithecus tattersalli" />
+        <species id="species:ncbi:48399" start="9043" end="9061" text="American crocodile" />
+        <species id="species:ncbi:48399" start="9064" end="9081" text="Crocodylus acutus" />
+        <species id="species:ncbi:184237" start="9089" end="9108" text="Morelet's crocodile" />
+        <species id="species:ncbi:37171" start="9419" end="9427" text="bontebok" />
+        <species id="species:ncbi:36500" start="9798" end="9805" text="S. Fish" />
+        <species id="species:ncbi:29073" start="9963" end="9974" text="polar bears" />
+        <species id="species:ncbi:29073" start="9977" end="9992" text="Ursus maritimus" />
+        <species id="species:ncbi:9708" start="10291" end="10305" text="Pacific walrus" />
+        <species id="species:ncbi:9708" start="10309" end="10336" text="Odobenus rosmarus divergens" />
+        <species id="species:ncbi:9778" start="10650" end="10666" text="Florida manatees" />
+        <species id="species:ncbi:9778" start="10669" end="10687" text="Trichechus manatus" />
+      </linnaeus>
+    </supplementary_information>
+  </commonNames>
   <places>
     <summaries>
       <contentlocation xmlns="http://wherein.yahooapis.com/v1/schema" xmlns:yahoo="http://www.yahooapis.com/v1/base.rng" xml:lang="en">
--- a/src/de/mpiwg/anteater/AnteaterController.java	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/AnteaterController.java	Mon Oct 22 14:21:14 2012 -0700
@@ -12,6 +12,8 @@
 import de.mpiwg.anteater.places.PlacesExtraction;
 import de.mpiwg.anteater.results.ResultController;
 import de.mpiwg.anteater.results.ResultsCarrier;
+import de.mpiwg.anteater.species.common.CommonNameFindController;
+import de.mpiwg.anteater.species.common.CommonNamesExtraction;
 import de.mpiwg.anteater.species.scientific.ScientificNameFindController;
 import de.mpiwg.anteater.species.scientific.ScientificNamesExtraction;
 import de.mpiwg.anteater.text.TextInformation;
@@ -39,6 +41,7 @@
 		ScientificNameFindController scienceNameFindController = new ScientificNameFindController(configuration);
 		PlaceFinderController placesController = new PlaceFinderController(configuration);
 		PersonFinderController personsController = new PersonFinderController(configuration);
+		CommonNameFindController commonNameFindController = new CommonNameFindController(configuration);
 		
 		List<TextInformation> textInformations = new ArrayList<TextInformation>();
 		TextManager textManager = new TextManager(configuration);
@@ -55,6 +58,10 @@
 			List<ScientificNamesExtraction> scienNameResults = scienceNameFindController.findScientificNamesInXML(info);
 			info.setScientificNamesExtractions(scienNameResults);
 			
+			//get common names and scientiric
+			List<CommonNamesExtraction> commonNamesResults = commonNameFindController.findCommonNamesInXML(info);
+			info.setCommonNamesExtractions(commonNamesResults);
+			
 			// get places
 			List<PlacesExtraction> placesResults = placesController.findPlacesInXML(info);
 			info.setPlacesExtractions(placesResults);
--- a/src/de/mpiwg/anteater/species/common/CommonNameFindController.java	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/species/common/CommonNameFindController.java	Mon Oct 22 14:21:14 2012 -0700
@@ -5,8 +5,12 @@
 import java.util.List;
 
 import de.mpiwg.anteater.AnteaterConfiguration;
+import de.mpiwg.anteater.species.common.impl.LinnaeusNameFinder;
+import de.mpiwg.anteater.species.scientific.IScientificNamesFinder;
 import de.mpiwg.anteater.species.scientific.ScientificNamesExtraction;
+import de.mpiwg.anteater.species.scientific.impl.GNRDNameFinder;
 import de.mpiwg.anteater.text.TextInformation;
+import de.mpiwg.anteater.text.TextPart;
 import de.mpiwg.anteater.xml.impl.AnalysisXMLManager;
 
 public class CommonNameFindController {
@@ -19,7 +23,7 @@
 		this.configuration = configuration;
 	}
 	
-	public List<ScientificNamesExtraction> findCommonNamesInXML(TextInformation info) {
+	public List<CommonNamesExtraction> findCommonNamesInXML(TextInformation info) {
 		List<CommonNamesExtraction> results = new ArrayList<CommonNamesExtraction>();
 		List<String> summaryAnalysisResults = new ArrayList<String>();
 		List<String> supplinfAnalysisResults = new ArrayList<String>();
@@ -40,6 +44,40 @@
 			configuration.getLogger().logMessage("found " + supplinfAnalysisResults.size() + " result(s).");
 		}
 		
+		ICommonNameFinder nameFinder = new LinnaeusNameFinder(configuration.getLogger());
+		
+		// if there are no results for summaries, ask GNRD name finding service.
+		if (summaryAnalysisResults.size() == 0) {
+			configuration.getLogger().logMessage(COMPONENT_NAME, "No results found for summaries, so will ask LinnaeusNameFinder.");
+						
+			for (TextPart sum : info.getSummaries()) {
+				String sumResult = nameFinder.findCommonNames(sum.getText());
+				if (sumResult != null) {
+					summaryAnalysisResults.add(sumResult);
+					
+					// if there is an analysis folder, add result to analysis file
+					if (analysisManager != null)
+						analysisManager.addSummaryCommonNamesResult(sumResult);
+				}
+			}
+		}
+		
+		// if there are no results for supplementary information, ask GNRD name fining service
+		if (supplinfAnalysisResults.size() == 0) {
+			configuration.getLogger().logMessage(COMPONENT_NAME, "No results found for supplementary information, so will ask LinnaeusNameFinder.");
+			
+			for (TextPart sInf : info.getSupplInfos()) {
+				String supinfResult = nameFinder.findCommonNames(sInf.getText());
+				if (supinfResult != null) {
+					supplinfAnalysisResults.add(supinfResult);
+					
+					// if there is an analysis folder, add result to analysis file
+					if (analysisManager != null)
+						analysisManager.addSupplInfCommonNamesResult(supinfResult);
+				}
+			}
+		}
+		
 		return null;
 	}
 	
--- a/src/de/mpiwg/anteater/species/common/impl/LinnaeusNameFinder.java	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/species/common/impl/LinnaeusNameFinder.java	Mon Oct 22 14:21:14 2012 -0700
@@ -2,12 +2,11 @@
 
 import java.io.InputStream;
 import java.util.HashMap;
-import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
-import java.util.logging.Logger;
 
-import martin.common.ArgParser;
+import org.apache.commons.lang3.StringEscapeUtils;
+
 import uk.ac.man.documentparser.dataholders.Document;
 import uk.ac.man.entitytagger.Mention;
 import uk.ac.man.entitytagger.doc.TaggedDocument;
@@ -48,9 +47,20 @@
 		Document doc = new Document("id", "title", "", text, text, Document.Text_raw_type.TEXT, "", null, Document.Type.OTHER, null, "", "", "", "", null);
 		TaggedDocument tagged = MatchOperations.matchDocument(matcher, doc);
 		List<Mention> species = tagged.getAllMatches();
-		for (Mention s : species)
-			System.out.println("found " + s.getMostProbableID() + ": " + s.getText() + " at " + s.getStart());
-		return null;
+		StringBuffer sb = new StringBuffer();
+		sb.append("<linnaeus>");
+		
+		for (Mention s : species) {
+			sb.append("<species id=\"" + StringEscapeUtils.escapeXml(s.getMostProbableID()) + "\" ");
+			sb.append("start=\"" + s.getStart() + "\" ");
+			sb.append("end=\"" + s.getEnd() + "\" ");
+			sb.append("text=\"" + StringEscapeUtils.escapeXml(s.getText()) + "\" ");
+			sb.append("/>");
+		}
+		
+		sb.append("</linnaeus>");
+		System.out.println(sb.toString());
+		return sb.toString();
 	}
 
 	public Postprocessor getPostprocessor(Map<String, String> comments,
--- a/src/de/mpiwg/anteater/text/TextInformation.java	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/text/TextInformation.java	Mon Oct 22 14:21:14 2012 -0700
@@ -4,6 +4,7 @@
 
 import de.mpiwg.anteater.persons.PersonsExtraction;
 import de.mpiwg.anteater.places.PlacesExtraction;
+import de.mpiwg.anteater.species.common.CommonNamesExtraction;
 import de.mpiwg.anteater.species.scientific.ScientificNamesExtraction;
 
 public class TextInformation {
@@ -13,6 +14,7 @@
 	private List<ScientificNamesExtraction> scientificNamesExtractions;
 	private List<PlacesExtraction> placesExtractions;
 	private List<PersonsExtraction> personsExtractions;
+	private List<CommonNamesExtraction> commonNamesExtractions;
 	private String filepath;
 	
 	public void setSummaries(List<TextPart> summaries) {
@@ -51,4 +53,10 @@
 	public List<PersonsExtraction> getPersonsExtractions() {
 		return personsExtractions;
 	}
+	public void setCommonNamesExtractions(List<CommonNamesExtraction> commonNamesExtractions) {
+		this.commonNamesExtractions = commonNamesExtractions;
+	}
+	public List<CommonNamesExtraction> getCommonNamesExtractions() {
+		return commonNamesExtractions;
+	}
 }
--- a/src/de/mpiwg/anteater/xml/impl/AnalysisXMLManager.java	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/xml/impl/AnalysisXMLManager.java	Mon Oct 22 14:21:14 2012 -0700
@@ -124,6 +124,14 @@
 	public void addSupplInfNamesResult(String result) {
 		addElementToDoc(result, "/analysis/scientificNames/supplementary_information");
 	}
+	
+	public void addSummaryCommonNamesResult(String result) {
+		addElementToDoc(result, "/analysis/commonNames/summaries");
+	}
+	
+	public void addSupplInfCommonNamesResult(String result) {
+		addElementToDoc(result, "/analysis/commonNames/supplementary_information");
+	}
 
 	private void addElementToDoc(String result, String xpath) {
 		IXMLParser parser = new JDOMParser(result, false);
--- a/src/de/mpiwg/anteater/xml/impl/templates/analysisFile.xml	Mon Oct 22 13:46:54 2012 -0700
+++ b/src/de/mpiwg/anteater/xml/impl/templates/analysisFile.xml	Mon Oct 22 14:21:14 2012 -0700
@@ -15,6 +15,14 @@
 			
 		</supplementary_information>
 	</scientificNames>
+	<commonNames>
+		<summaries>
+			
+		</summaries>
+		<supplementary_information>
+			
+		</supplementary_information>
+    </commonNames>
 	<places>
 		<summaries>